一种基于遗传算法的多模型预测控制方案

针对时滞系统的特点,提出一种基于遗传算法的多模型预测控制方法.利用多个模型逼近整个被控对象的动态特性,针对每个子模型,设计相应的子控制器,通过加权的方式,将有限个控制量映射成最终的控制量;子控制器采用预测控制算法,利用遗传算法得到权值的全局最优解;比较该控制方案和PID控制在时滞系统中的应用效果.仿真结果表明：该算法不但克服了建模的不精确性和参数的时变性,使系统具有良好的控制性能,而且具有较强的鲁棒性.

Multiple Models Predictive Control Based on Genetic Algorithm

In terms of the characteristics of time lag system,a method of multiple models predictive control based on genetic algorithm is proposed. Multiple models approach the dynamic character of the controlled object. Design corresponding controller to each model. By means of weighting, get final controlled variable with the limited controlled variable. Each controller adopts predictive control method. Get the global optimum by genetic algorithm. Compare the effect of PID in time lag system. Simulation result indicates the method not only can overcome inaccuracy of modeling and time variation of parameters,but also have good control performance and stronger robustness.